Papers by Kyungmin Kim

8 papers
Fooling the LVLM Judges: Visual Biases in LVLM-Based Evaluation (2025.emnlp-main)

Copied to clipboard

Challenge: a recent study has found that large vision–language models are vulnerable to visual biases that inflate scores without altering semantic content.
Approach: They propose a novel meta-evaluation benchmark that exhibits diverse score distributions.
Outcome: The proposed model exhibits vulnerability across all domains, and combines multiple biases amplifies their effects, and pairwise evaluations are similarly susceptible.
Mitigating Hallucinations in Large Vision-Language Models via Summary-Guided Decoding (2025.findings-naacl)

Copied to clipboard

Challenge: Large Vision-Language Models (LVLMs) generate detailed and coherent responses from visual inputs but are prone to generate hallucinations due to an over-reliance on language priors.
Approach: They propose a method that reduces the text context and controls only the image-related POS tokens to maintain text quality by reducing the text contextualization.
Outcome: The proposed method achieves state-of-the-art performance on object hallucination benchmarks and achieves Pareto optimality among the existing methods.
Selective Perception: Learning Concise State Descriptions for Language Model Actors (2024.naacl-short)

Copied to clipboard

Challenge: Recent large language models support longer contexts, but requiring them to process redundant or irrelevant data increases inference time and cost.
Approach: They propose a method that leverages a small finetuned LM to sample the minimal set of input features that maximizes the performance of a downstream LM.
Outcome: The proposed method reduces the length of LM actor input by 87% and 99% while improving task success rates by 158% and 54% on NetHack and robot planning.
ChronoBias: A Benchmark for Evaluating Temporal Group Bias in the Time-sensitive Knowledge of Large Language Models (2025.findings-emnlp)

Copied to clipboard

Challenge: Using a template-based semi-automated generation method, we evaluate time-conditional group bias in time-sensitive knowledge of large language models (LLMs).
Approach: They propose a template-based semi-automated generation method to construct a time-conditional group bias benchmark.
Outcome: The proposed method balancing quality-quantity trade-off in existing benchmark curation approaches.
Reliability-Aware Adaptive Self-Consistency for Efficient Sampling in LLM Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Self-consistency improves reasoning reliability but incurs substantial inference cost . Adaptive self-consistent methods rely on count-based stopping rules that treat all responses equally .
Approach: They propose a method that reframs adaptive sampling from response counting to evidence sufficiency by leveraging response-level confidence.
Outcome: The proposed method reduces inference cost by up to 70% while preserving accuracy on GSM8K.
On Sample-Efficient Code Generation (2023.emnlp-industry)

Copied to clipboard

Challenge: Existing approaches to code generation rely on rejection sampling to generate multiple code snippets then select the best.
Approach: They propose a framework that prioritizes sampling on test problems that models can solve.
Outcome: The proposed framework reduces sampling costs while maintaining comparable code generation performance.
Pivotal Role of Language Modeling in Recommender Systems: Enriching Task-specific and Task-agnostic Representation Learning (2023.acl-long)

Copied to clipboard

Challenge: Recent studies have proposed unified user modeling frameworks that leverage user behavior data from various applications.
Approach: They propose to use user behavior sequences as plain text to represent rich information in any domain or system without losing generality.
Outcome: The proposed frameworks achieve excellent results on diverse recommendation tasks and can be used on unseen domains and services.
Contrastive Learning as a Polarizer: Mitigating Gender Bias by Fair and Biased sentences (2024.findings-naacl)

Copied to clipboard

Challenge: Recent studies have highlighted social biases inherent in training data can lead models to learn and propagate them.
Approach: They propose a contrastive learning method that uses anchor points to push further negatives and pull closer positives within the representation space.
Outcome: The proposed method achieves state-of-the-art in the ICAT score on the StereoSet, a benchmark for measuring bias in models.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations